Class-Incremental Learning on Tiny-ImageNet Split
25.15AccuracyMANGO
Evaluation Results
| Method | Links | |||
|---|---|---|---|---|
| MANGOBuffer=2k2026.05 | 25.15 | 23.45 | 17.44 | |
| MANGOBuffer=4k2026.05 | 24.73 | 23.33 | 17.27 | |
| LODEBuffer=4k2026.05 | 11.68 | 15.74 | 11.64 | |
| LODEBuffer=2k2026.05 | 9.97 | 13.96 | 9.12 | |
| DER++Buffer=4k2026.05 | 6.9 | 12.95 | 7.47 | |
| DER++Buffer=2k2026.05 | 6.62 | 12.05 | 6.73 | |
| ERBuffer=4k2026.05 | 5.93 | 12.02 | 6.62 | |
| ER-ACEBuffer=4k2026.05 | 5.93 | 12.02 | 6.62 | |
| ERBuffer=2k2026.05 | 5.1 | 4.86 | 2.11 | |
| ER-ACEBuffer=2k2026.05 | 5.1 | 10.75 | 4.83 | |
| LUCIRBuffer=2k2026.05 | 3.05 | 8.14 | 4.07 | |
| LUCIRBuffer=4k2026.05 | 2.84 | 8.27 | 4.19 | |
| GDUMBBuffer=4k2026.05 | 2.68 | 8.23 | 6.09 | |
| FTBuffer=—2026.05 | 1.94 | 4.86 | 2.11 | |
| GDUMBBuffer=2k2026.05 | 1.58 | 5.68 | 4.14 | |
| iCaRLBuffer=4k2026.05 | 0.64 | 2.3 | 1.68 | |
| iCaRLBuffer=2k2026.05 | 0.61 | 1.86 | 1.42 |